Ok, Mikey, so you are the culprit hogging all the GR#1 tasks. Gimmmmeeee!! Hah ha hah, LOL
I just looked and I have 400+ grp#1 in the queue on both of my systems. It is a November miracle :)
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
I just tried again earlier today on one host and got up to around 50 tasks at one time. Left it to its own devices while I vacuumed the house and just now looked and I had 1 GR#1 task running and the gpus were doing MW to keep busy.
So back to the work venue for both GR#1 and BRP7 which is the only setup that keeps Einstein on my gpus.
How do I get BRP7 WUs? My project prefs are set to the 2 that specify GPU only (I don't want any CPU work). One of my machines has ~400 wus onboard (normal) , the other is flirting with 0 for the last 3-4 days. If I check off MEERKAT, will that do it, or is there something else I need to do?
So back to the work venue for both GR#1 and BRP7 which is the only setup that keeps Einstein on my gpus.
Just switched my Epyc box back to Grp#1 and brp7.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Yes, MeerKat in the project preferences is for the BRP7 gpu tasks.
And assume the drivers are up to date enough to run the tasks. The Boinc Log will notify you if it is unwilling to download brp7 tasks.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Ok, Mikey, so you are the culprit hogging all the GR#1 tasks. Gimmmmeeee!! Hah ha hah, LOL
I just looked and I have 400+ grp#1 in the queue on both of my systems. It is a November miracle :)
Half a miracle. The Epyc seems to have run out of grp#1 tasks. It munching Milky way.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
My three most powerful rigs have run out of E@H work and are now crunching on behalf of Amicable Numbers. My account is now associated with 5,793 E@H FGRPB1G tasks in progress, which is approximately only 1.2 days of work and is significantly smaller than the 10,058 E@H FGRPB1G tasks associated with my account approximately two weeks ago.
On the other hand, my account is associated with almost 3% of the total FGRPB1G tasks in progress (214,360), so I am not complaining, particularly since I employ more or less obsolete hardware kluged in a manner that defies common (and financial) sense.
I will monitor the total FGRPB1G tasks in progress during the next week or two to see if the shortage of FGRPB1G tasks improves.
Again, not complaining - just reporting.
"I was born in a small town, and I live in a small town." - John Mellencamp
Keith - thanks for that. I finally got around to adding MK to my list of OK apps to run this am (I had my gall bladder out at the end of October - not recommended for a fun holiday). and they seem to be doing fine, at least on my machine (Win7Relic) that was flirting with 0 tasks for the last week or so.
A question:
I noticed that when running only the GR app I had (on both machines) Task Mgr (Win 7) saying that CPU usage was around 40% or so and kernel time ~1-2%. Now Win7Relic is showing ~30% cpu and ~20% kernel time.
(I assume because GR is openCL and MK is cuda). Also, GR WUs are 99+% CPU time and MK are 50-65% CPU (in BoincTasks). Would that imply that going down from 3 WU/GPU to 2 or 1 would get better performance?
Wasn't true under opencl (I tested that before), but maybe CUDA is so much less efficient at task switching on the GPU that maybe it would pay to drop down the parallel processing? Just a guess - oops, I mean, hypothesis.
Keith Myers wrote: Ok,
)
I just looked and I have 400+ grp#1 in the queue on both of my systems. It is a November miracle :)
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
I just tried again earlier
)
I just tried again earlier today on one host and got up to around 50 tasks at one time. Left it to its own devices while I vacuumed the house and just now looked and I had 1 GR#1 task running and the gpus were doing MW to keep busy.
So back to the work venue for both GR#1 and BRP7 which is the only setup that keeps Einstein on my gpus.
How do I get BRP7 WUs? My
)
How do I get BRP7 WUs? My project prefs are set to the 2 that specify GPU only (I don't want any CPU work). One of my machines has ~400 wus onboard (normal) , the other is flirting with 0 for the last 3-4 days. If I check off MEERKAT, will that do it, or is there something else I need to do?
Thanks!
Jon
Yes, MeerKat in the project
)
Yes, MeerKat in the project preferences is for the BRP7 gpu tasks.
Project Preferences
Keith Myers wrote: So back
)
Just switched my Epyc box back to Grp#1 and brp7.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Keith Myers wrote: Yes,
)
And assume the drivers are up to date enough to run the tasks. The Boinc Log will notify you if it is unwilling to download brp7 tasks.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M wrote: Keith Myers
)
Half a miracle. The Epyc seems to have run out of grp#1 tasks. It munching Milky way.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Keith Myers wrote: Ok,
)
You know I would if I could!! :-))
My three most powerful rigs
)
My three most powerful rigs have run out of E@H work and are now crunching on behalf of Amicable Numbers. My account is now associated with 5,793 E@H FGRPB1G tasks in progress, which is approximately only 1.2 days of work and is significantly smaller than the 10,058 E@H FGRPB1G tasks associated with my account approximately two weeks ago.
On the other hand, my account is associated with almost 3% of the total FGRPB1G tasks in progress (214,360), so I am not complaining, particularly since I employ more or less obsolete hardware kluged in a manner that defies common (and financial) sense.
I will monitor the total FGRPB1G tasks in progress during the next week or two to see if the shortage of FGRPB1G tasks improves.
Again, not complaining - just reporting.
"I was born in a small town, and I live in a small town." - John Mellencamp
Keith - thanks for that. I
)
Keith - thanks for that. I finally got around to adding MK to my list of OK apps to run this am (I had my gall bladder out at the end of October - not recommended for a fun holiday). and they seem to be doing fine, at least on my machine (Win7Relic) that was flirting with 0 tasks for the last week or so.
A question:
I noticed that when running only the GR app I had (on both machines) Task Mgr (Win 7) saying that CPU usage was around 40% or so and kernel time ~1-2%. Now Win7Relic is showing ~30% cpu and ~20% kernel time.
(I assume because GR is openCL and MK is cuda). Also, GR WUs are 99+% CPU time and MK are 50-65% CPU (in BoincTasks). Would that imply that going down from 3 WU/GPU to 2 or 1 would get better performance?
Wasn't true under opencl (I tested that before), but maybe CUDA is so much less efficient at task switching on the GPU that maybe it would pay to drop down the parallel processing? Just a guess - oops, I mean, hypothesis.